- Email: [email protected]
Targeted badger removal and the subsequent risk of bovine tuberculosis in cattle herds in county Laois, Ireland
Preventive Veterinary Medicine 88 (2009) 178–184
Contents lists available at ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Targeted badger removal and the subsequent risk of bovine tuberculosis in cattle herds in county Laois, Ireland F.J. Olea-Popelka a,e,*, P. Fitzgerald b, P. White c, G. McGrath c, J.D. Collins c, J. O’Keeffe c,d, D.F. Kelton a, O. Berke a, S. More c, S.W. Martin a a
Department of Population Medicine, Ontario Veterinary College, University of Guelph, Clinical Research Building, Guelph, Ont., Canada N1G 2W1 District Veterinary Office County Laois, Ireland Centre of Veterinary Epidemiology and Risk Analysis, University College Dublin, Belfield, Ireland d Department of Agriculture, Fisheries and Food, Agriculture House, Kildare Street, Dublin 2, Ireland e Animal Population Health Institute, Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins 80523, USA b c
A R T I C L E I N F O
A B S T R A C T
Article history: Received 24 July 2007 Received in revised form 3 September 2008 Accepted 8 September 2008
We investigated the impact of targeted removal of badgers on the subsequent bovine tuberculosis (BTB) risk in cattle herds in county Laois, Ireland. The study period was 1989– 2005. For each of 122 targeted badger-removal licenses (permit to remove badgers in the proximity of cattle herds undergoing a serious BTB episode), the herd number (index herd) for which the license was given was obtained. The herds in the proximity of the index herd were identified from another database. The main ‘‘exposure’’ in our study was the geographical location of herds relative to the area in which targeted badger removal was conducted. We categorized herds into five different exposure groups: herds were classified as non-exposed and denoted as group 0 (reference group) if they were located 500 m or more from the edge of any parcel of land of the index herd; group 1, was the index herds, group 2 the immediate (contiguous) neighbors of the index herd, group 3 herds were not immediate neighbors but within 150 m and group 4 herds were between 150 m and 500 m distance from the edge of any parcel of land of the index herd, respectively. We conducted a survival analysis (allowing multiple failures per herd) to compare the hazard of having a BTB episode in any of the four groups of exposed herds vs. the hazard in herds in the reference group. We controlled for other known risk factors as well taking into account a temporal component. Our analysis showed that the hazard ratio for the index herds (group 1) were non-significantly increased, indicating that there was no difference in the hazard of failing a BTB test (after the targeted badger removal was conducted) between index herds and reference herds. For the rest of the herds farther away from badger removal activities the hazards were lower than herds in areas not under badger removal. The hazard in the reference group decreased over the study period. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Bovine tuberculosis Badger removal Ireland
1. Introduction
* Corresponding author at: Animal Population Health Institute, Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins 80523, USA. Tel.: +1 970 2975064; fax: +1 970 4912250. E-mail address: [email protected] (F.J. Olea-Popelka). 0167-5877/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.prevetmed.2008.09.008
One of the current constraints to the eradication of bovine tuberculosis (BTB) in Ireland is the existence of an important wildlife reservoir for Mycobacterium bovis (M. bovis); namely the badger (Meles meles). In Ireland, several studies have provided evidence that tuberculous badgers
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
are associated with M. bovis infection in cattle (O’Connor and O’Malley, 1989; Dolan, 1993; Martin et al., 1997). In addition, two field trials known as the ‘‘East Offaly’’ and ‘‘Four Area’’ projects (O’Mairtin et al., 1998; Eves, 1999; Griffin et al., 2005) have shown that intensive badger removal lead to a decrease in the subsequent levels of tuberculosis in cattle herds in the areas in which the badgers were removed compared to areas with restricted targeted, or no, removal of badgers. Since the early 1990s, targeted badger removal (also known as ‘‘reactive’’ badger culling) has been conducted through Ireland. This practice is initiated when a veterinary inspector, in the course of an epidemiological investigation of a BTB breakdown strongly suspects badgers as the source of infection. After obtaining a license to capture badgers, from the National Parks and Wildlife Service of the Department of the Environment, Heritage and Local Government, an area of 1–2 km radius around the affected farm is surveyed and badger removal takes place until few badgers remain and/or the BTB problem in the area has been cleared. Despite the potential large size of the removal area, usually only those setts on or close to the index herd have badgers removed. This activity differs from the two badger-removal projects mentioned above in the sense that badgers are removed only around herds with a serious BTB problem (i.e. two or more cattle deemed as ‘‘standard reactors’’ and one animal with a BTB lesion). Reactive culling is a response to the problem in the area and is an attempt to prevent the spread of BTB from badgers to other herds in the vicinity of the affected (index) herd. The objective of our study was to assess the impact of ‘‘targeted’’ badger removal on the subsequent BTB risk of herds in one county, Laois, in Ireland. For this, we conducted a survival analysis in which the outcome was ‘‘time to a new breakdown of BTB’’. Herds were classified as having different levels of exposure to badgers in the removal area based on their geographical location relative to the index herd. We tested the hypothesis that the hazard of BTB for herds in areas in which targeted badger removal was conducted, and in surrounding areas, does not differ from the hazard of BTB for herds in areas in which targeted badger removal was not conducted. 2. Material and methods The study was conducted in county Laois, Ireland. County Laois is a centrally located inland county, 90 km southwest of the capital city Dublin. All badger removal licenses files were collected for the period 1990–2005 from the Laois District Veterinary Office (DVO). Data on the herd (index herd) for which the license was issued, the date at which the license was opened, the date of the last badger removal (end of the license), the number of badgers removed, and the number of tuberculous badgers removed per license was entered into a Excel spread sheet (Microsoft1 Office, 2003). If a license was reopened we included this as a new license. The herd BTB history data were obtained from the Centre for Veterinary Epidemiology and Risk Assessment at the University College Dublin, Ireland. This centre
179
maintains computerized data on all herd-test results for BTB since January 1989. In Ireland, all herds are tested at least once annually; herds suspected of having BTB, or herds with BTB, are tested more frequently. Data were obtained for all herds in county Laois between the first test record (on or after 1 January 1989) until the last test record during the year 2005. 2.1. Exclusion criteria All herds containing fewer than seven animals were excluded from our study. Also, herds that ceased their activities before 2003 were not included in the study. This exclusion was necessary because the computerized ‘‘neighboring’’ file (see below) which was essential for our study contained data only on herds that were active during 2003. 2.2. Exposure categories Herd exposure was based on their location relative to the index herd (and hence relative to the activity of targeted badger removal). Using data from the ‘‘neighboring’’ file created with a Geographical Information System (GIS), herds that were located in areas 500 m or more from the edge of any parcel of land of the index herd at the time of the BTB breakdown were classified as non-exposed and denoted as group 0 (reference group). Herds that originated a license were identified as the index-herds (group = 1). All the immediate neighbors (<25 m distance from the edge of any parcels of land of the index herd) were identified and classified as group 2 herds. Herds at distances of 25 < 150 m and 150 < 500 m distances of the edge of any parcel of land of the index-herd at the time of the BTB breakdown were identified as exposure groups 3 and 4, respectively. Considering that typical Irish farms are composed of several parcel of land, and each parcel of land varies in their dimensions, herds recorded at more than 500 m of any of the edge of the index farm could be located at distances as large or larger than 3 km (O’Keeffe, 2007 personal communication). Thus, the distance measures used to create the exposure categories are ordered but the cutpoints do not reflect the actual distance from the badger removal areas. Herd classification was allowed to change over time as explained below. 2.3. Exposure over time The starting point to accumulate time at risk was the first recorded herd test in 1989, or the first herd-test of new herds after that date. If the herd was positive to BTB at its first test, we used the date at which the herd was declared ‘‘clear’’ (i.e. after the herd had passed three consecutive negative herd-tests, usually a period of approximately 10 months) to begin accumulating time-at-risk. Assuming the herd was not exposed to badger removal at its first test, herds accumulated time at risk in exposure group 0 until one of the following scenarios occurred: the herd experienced a change in exposure status (to one of the groups 1–4) (Fig. 1), or
180
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
Fig. 1. Exposure over time: change in exposure categories for herds in county Laois, Ireland after the targeted badger removal was ended, 1989–2005.
Fig. 2. (a) Exposure over time: example of a herd tested for the last time (censored time) without any events during the study period in county Laois, Ireland 1989–2005. (b) Exposure over time: example of the progress herds could have from what we considered as minimally exposed (group 4) to higher levels of exposure (groups 3, 2, or 1) for the duration of the study 1989–2005 in county Laois, Ireland.
the herd experienced a BTB episode (failure time) (Figs. 1 and 3), or the herd was tested for the last time (censored time) (Fig. 2a). When a herd became an index herd (group 1), it’s exposure classification remained as a group 1 for the entire duration of our study. The classification of immediate neighbors (group 2) did not change unless they became an index herd. Herds in groups 3 or 4 only changed their status if they subsequently were located closer to another index herd (e.g. from being >150–500 m. distant from an index herd (group 4), to being an immediate neighbor for another index herd (group 2). Herds could only progress from what we considered as minimally exposed (group 4) to higher levels of exposure (groups 3, 2, or 1) for the duration of the study (Fig. 2b). With the exception of the reference group, exposure to the badger removal activities (group 1 through 4) was deemed to begin when badger removal activities ended. This rule was invoked because badger licenses often were kept open until new tuberculosis problems in the groups 1 and 2 herds had been overcome.
2.4. Outcome The outcome of experiencing a herd breakdown with tuberculosis denoted a herd having at least one or more standard reactors1 at the Single Intradermal Comparative Tuberculin Test (SICTT). If a herd developed a BTB episode, it was removed from the at-risk group in its existing exposure category until it had passed three negative herd tests. Thereafter, it was deemed to be at risk of another BTB breakdown; thus a herd could have multiple failures (BTB episodes) during the study period (Fig. 3 and Table 2). The cumulative exposure time in each exposure category was calculated until the herd broke down with BTB (Fig. 3). The censoring time was the date of the last recorded herd test before 1 January 2006.
1 In Ireland, a standard reactor is recorded when the bovine reaction is >4 mm larger than the avian reaction, or if local clinical signs (such as edema, exudation, necrosis or pain) are present at the site of injection of bovine tuberculin (Monaghan et al., 1994).
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
181
Fig. 3. Outcome over time: example of a herd having multiple failures (BTB episodes) during the study period 1989–2005 in county Laois, Ireland.
2.5. Statistical analysis SAS V8.2 (Statistical Analytical System Institute Inc., Cary, NC, USA), was used for data manipulation, and descriptive analysis. STATA (Stata Statistical Software: Release 9 College Station, TX, USA) was used for further data manipulation and survival analysis. Univariable and multivariable Cox’s proportionalhazard models were used to contrast the hazards for a BTB breakdown in each of the four exposure groups relative to the hazard in the non-exposed group (reference group) using STCOX in STATA. We included the herd size, herd type, the length of the badger removal period, the number of badgers and tuberculous badgers removed and the time (year) in which the removal activities took place as covariates. We accounted for clustering (repeated measurements within a herd) by using the ROBUST standard error option. We checked the proportional-hazard assumption by assessing, visually, if the log( log) survival lines were parallel for the different exposure groups (Allison, 1995). More formally, we assessed if the hazard-ratio varied over time by analyzing the x2 Shoenfeld residuals (Grambsch and Therneau, 1994), and if significant (p < 0.05), we included the exposure as a time varying covariate (Dohoo et al., 2003). Exposure time and time-to-failure were measured in months. Interaction terms between the potential confounders and the exposure groups were evaluated. 3. Results 3.1. Descriptive Over the 16-year study period, 135 badger removal licenses were granted; complete data were available for 122 of these. The data for nine of the omitted licenses were excluded because of the absence of the index herd in the 2003 contiguous file, making it impossible for us to locate their neighbors; four others for which the end date of the licence was not recorded were also omitted. In total, 1257 badgers were removed and 15% of these (n = 182) were deemed to have BTB lesions at the gross post-mortem examination. Table 1 shows the distribution of badgers and tuberculous badgers removed at the completion of each license by year. There was no statistically significant
difference in the annual risk of badgers with a gross BTB lesion (x2 = 13.0, d.f. = 12, p = 0.36). Neither was there a significant linear trend on the levels of badger BTB over time using the Cochran-Armitage test for trend (x2 = 1.531, d.f. = 1, p-value = 0.216). On average 3645 herds were tested each year. We included only data on 2904 herds after implementing our exclusion criteria. Among these herds 55% did not have a BTB breakdown over the study period. The distribution of number of BTB breakdowns per herd is shown in Table 2. Among herds that had BTB, 36% had repeated episodes; the distribution of occurrences fitted a negative binomial but not a Poisson distribution. The 1st quartile, median and 3rd quartile for the duration of the BTB episode were 11, 12 and 15 months, respectively. Table 1 Number of badgers and tuberculous badgers removed by year in county Laois, Ireland 1991–2004. Year
No. of badgers
No. of tuberculous badgers
% Positive
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
136 62 18 48 59 202 114 159 34 69 114 161 59 22
21 10 1 6 6 34 24 25 3 6 18 24 4 0
15.4 16.1 5.6 12.5 10.2 16.8 21.1 15.7 8.8 8.7 15.8 14.9 6.8 0.0
Table 2 Number of BTB episodes per herd in county Laois, Ireland 1989–2005. Number of BTB breakdowns
Number of herds
0 1 2 3 4 5
1590 846 325 111 28 4
Total
2904
Percentage 54.8 29.1 11.2 3.8 1.0 0.1 100
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
182
Table 3 Distribution of time at risk (in months) by exposure category for herds in county Laois, Ireland 1989–2005. Percentiles
Exposure categorya 0
1
2
3
4
Smallest 1% 25% 50% 75% 99%
0.2 3.8 35.4 83.4 156 202.8
0.2 0.3 24.8 53 94.1 172
0.03 1.3 37.2 66.7 108.6 154.8
0.1 0.9 36.7 61.2 104.2 176.7
0.03 0.8 34.3 60.8 103.8 174.8
Largest
208.1
178
188.2
190.5
181.3
a
See Section (2.2) for description of each exposure category.
hazard for the index herds did not differ from the referent. The hazard ratio was not constant for herds in groups 2 and 3 (p 0.001 and p = 0.043, respectively); the hazard ratio for these groups was initially low, but increased slightly over the years while remaining low relative to the hazard for herds in group 0. Exposure group 4 herds had a lower risk of a BTB breakdown than the referent group. As the number of animals tested in a herd increased by 10, so did the hazard of failing a BTB herd test (HR = 1.05, p < 0.001). The highest hazard for year in 1990 (HR = 2.2, p < 0.001) significantly decreased over time until the end of 1995. The hazards between 1996 and 2003 did not differ from that in 1989 (data not shown). 4. Discussion
Historical records for county Laois showed that the herd incidence (% of new herds restrictions) over the 16-year study period decreased beginning in 1991 and again in 1996 but retained a constant level from 1999 onwards. The distribution of time at risk (in months) for each exposure category is shown in Table 3; not surprisingly, exposure category ‘‘0’’ had the longest median (or mean) values due to the time sequence of initiating badger removal licenses during the study period. 3.2. Survival analysis All the exposure categories and confounders had a significant association with the time to fail a BTB test in a simple univariate survival model (Table 4). The results for the main risk factor, exposure category, indicated that group 1 herds had a higher hazard than group 0 herds. Conversely, exposure groups 2, 3 and 4 had lower hazard ratios indicating longer survival. As the number of badgers and tuberculous badgers removed increased, the hazard of a herd failing a BTB test decreased. As the number of animals tested in a herd increased, so did the hazard of failing a herd test. The hazard for beef and suckler2 herds was lower than dairy herds. The hazard of failing a herdtest decreased over the study period. After controlling for the other variables in the model, herd type, the effect of number of badgers removed and tuberculous badgers removed, was no longer significant. The log( log of the survival probability) graph for a BTB breakdown, by exposure category, revealed that the survival lines were not parallel, visually, indicating that the hazard ratio for some exposure categories varied significantly over time. The Shoenfeld residuals test also indicated that our model violated the proportionality assumption (x2 = 212.8, d.f. = 22, p < 0.001). Thus, we controlled and adjusted for this by including the exposure categories as a time varying covariate (Dohoo et al., 2003). The final multivariable survival model is shown in Table 5. We controlled for year of study and herd size. The
2 Breeding cows from the beef type breeds (Charolais, Limousine, Hereford, Aberdeen Angus, etc.) crossed with a beef breed type bull to produce animals suitable for production of high-quality beef. Calves when born run with their dam for 4–8 months. A suckler herd normally does not produce a finished beef animal, but rather a yearling animal for further feeding.
Since 2003, a considerable amount of controversy has surrounded the issue of badger removal as a measure for controlling BTB in cattle. The East Offaly and Four Area projects (Eves, 1999; Griffin et al., 2005) compared areas in which extensive proactive badger removal was conducted to eliminate badgers as a potential source of M. bovis infection for cattle, and this resulted in a decrease of the BTB levels in cattle relative to those where minimal (restricted targeted), or no, badger removal took place (Eves, 1999; Griffin et al., 2005). In contrast, in the randomized badger culling trial (RBCT) in southwest England, areas with ‘‘reactive’’ badger culling had an increased incidence of BTB in cattle (Donnelly et al., 2003) and herds in areas around the extensive badger removal areas also had an increased incidence of BTB. Badger social disruption was suggested as the main reason for the observed increased incidence of BTB in cattle-herds in these areas in England. In our study, the protective hazard ratios obtained for groups 2–4 (0.42, 0.33 and 0.48, respectively) indicate that herds in areas around the targeted removal activities had a longer survival time to the next BTB episode compared to herds outside of these areas (reference group). One could argue that, the lower hazard for the exposed groups (2–4) could be an ‘‘artifact’’ due to an increase of BTB in the nonexposed group more than a decrease of BTB in the exposed groups. We investigated this aspect, but found no evidence of an increase over time of the BTB levels in herds in areas in which badgers were not removed. More specifically, we observed a statistically significant decrease of the hazard among herds in the reference group over the first 6 years of the study, after which it remained constant (data not shown). Thus, the lower hazard ratios for herds in groups 2–4 suggest a true decrease of the risk of BTB among these herds. With regards to potential misclassification bias in our outcome variable (BTB event), we recognize that the SICTT is not a perfect test. As a reasonable guide, the SICTT as administered in Ireland has a specificity (proportion of non-infected cattle that test negatively) of 99.8–99.9% and a sensitivity (proportion of infected cattle that test positively) of between 68% and 95% (Monaghan et al., 1994). Thus, it is likely that at the herd test, not all animals infected with M. bovis are detected. Nonetheless, herd sensitivity is higher than the animal-level sensitivity and
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
183
Table 4 Univariate hazard ratios, of future bovine tuberculosis herd breakdown after targeted badger removal took place in county Laois, Ireland 1989–2005. Variables
Hazard ratio
Robust S.E.
P-value
95% HR CL
Reference 1.62 0.71 0.48 0.57
– 0.18 0.05 0.08 0.05
– <0.0001 <0.0001 <0.0001 <0.0001
– 1.29, 0.63, 0.36, 0.48,
a
1. Badger removal exposure category 0 1 2 3 4
2.04 0.82 0.63 0.69
2. Number of badgers removed (increase in 1 badger)
0.97
0.005
<0.0001
0.97, 0.98
3. Number of tuberculous badgers removed (increase in 1 positive badger)
0.90
0.02
<0.0001
0.85, 0.94
4. Herd size (Increase in 10 cattle)
1.05
0.008
<0.0001
1.04, 1.07
5. Length of removal (months)
0.99
0.0001
0.018
0.99, 0.99
6. Year of study period (increase in 1 year)
0.97
0.005
<0.0001
0.96, 0.99
7. Herd type Beef Suckler
0.66 0.72
0.046 0.043
<0.0001 <0.0001
0.57, 0.75 0.64, 0.81
a
See Section (2.2) for description of each exposure category.
Table 5 Multivariable hazard ratios by exposure group, of future bovine tuberculosis herd breakdown after targeted badger removal took place in county Laois, Ireland 1989–2005. Variables
Hazard ratio
Robust S.E.
P-value
95% HR CL
Badger removal exposure category 0 1 2 3 4
Reference 1.15 0.42 0.33 0.48
– 0.25 0.06 0.08 0.08
– 0.51 <0.0001 <0.0001 <0.0001
– 0.74, 0.33, 0.21, 0.35,
Covariates Herd size (inc. in 10 animals)
1.05
0.007
<0.0001
1.04, 1.07
Time varying covariates (by month) 1 2 3 4
1.004 1.009 1.007 1.004
0.0037 0.0017 0.0035 0.0022
0.32 <0.0001 0.043 0.107
0.99, 1.01, 1.00, 0.99,
a
a
1.78 0.55 0.54 0.65
1.01 1.01 1.01 1.01
See Section (2.2) for description of each exposure category.
should be non-differential with respect to major exposure factors. If a herd had a BTB event during the study period, we considered that herd as ‘‘clear’’ (hence, cumulating ‘‘time at risk’’ again) only if the herd tested negatively on at least three consecutive tests (2 reactor re-test plus one ‘‘6month’’ check test). This was a ‘‘conservative’’ approach and should minimize any bias. In this study, we focused on the herd-risk of BTB in areas with ‘‘targeted’’ badger removal, analogous to those areas described as ‘‘reactive’’ culling in other studies, in comparison to the herd-risk in areas that were relatively distant from badger removal activities. We selected county Laois for our study because of the availability of welldocumented data on targeted badger removal over a considerable period of time (since 1989 onwards). Historical data show that county Laois over the last 10 years has experienced a decrease in BTB levels. In our study, the percentage of tuberculous badgers was obtained only by post-mortem BTB gross lesion findings. The percentage of tuberculous badgers removed did not differ over the study period. We acknowledge that the
post-mortem diagnosis it is not a sensitive method for detecting tuberculosis. Corner et al., 2005 have described the true level of BTB infection among badger population with a more sensitive methodology for identifying infection. The results of such studies have shown that approximately 44% of badgers are infected with M. bovis. The lower sensitivity in our study should be nondifferential by exposure category, but could reduce the significance of this specific variable. Indeed, the number of tuberculous badgers removed was not a predictor of future hazard of BTB in cattle herds. The assigning of a badger-removal exposure category to a herd was done by considering the end date for each badger removal license as a starting point of the time at risk for each exposure category. We did this, keeping in mind that badgers tended to be removed around groups 1 and 2 herds until the BTB problem in cattle herds had been cleared. During this time, these herds often were tested more frequently than herds in other areas, and hence would have had an apparent (if not real) increased risk of BTB.
184
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
In our final model, the non-significant hazard ratio obtained for the index herds (group 1), indicated that there was no difference in the hazard of failing a BTB test (after the targeted badger removal was conducted) between index herds and herds in the reference group (at considerable distances from badger removal activities). Previous research indicated that the group of index herds would be classified as a future ‘‘high’’ risk group of herds, considering they recently had a serious BTB episode (Olea-Popelka et al., 2004). Data on badgers, and badger removal, were not available for that study. In the current study, herds that had a severe BTB breakdown did not have a higher risk of failing a future BTB test after the badger removal was completed in the area suggesting a beneficial impact of targeted removal on the survival time in these herds. The index herd and its immediate neighbors undergo a more frequent testing regime than other herds. As conducted in Ireland, this strategy increases the probability of detecting infected animals (therefore herds) and together with the removal of badgers, decreases the reservoirs of infection for other cattle (herds) in the area. For this reason, we decided to accumulate time at risk for each exposure category only after the badger removal activities ceased. In addition, we assumed only a monotonic inverse association of exposure category with BTB risk and did not allow for other possibilities. We consider it important to understand the impact of badger removal activities on the dynamics of badger populations, and to distinguish between social disruption of badger groups and the effects of removal on levels of BTB in cattle herds. Alternative analyses of the data to evaluate these assumptions are planned in the future. Any causal link between reactive badger removal and bovine tuberculosis in cattle herds is certain to require a time lag sufficient to enable the hypothesized cascade of events to be completed including perturbation in disturbed badger populations, increased movement and contact between badgers within and without the same social groups, increased contact between badgers and cattle, infection of cattle, the development of responsiveness to tuberculin following establishment of infection in cattle under natural conditions, the scheduling of the (annual) herd test, the inspection of reactor animals at slaughter and laboratory confirmation of disease (More, 2007 personal communication). The risks posed by tuberculous badger to cattle have being recently described and analyzed by Corner, 2006. After considering different factors, Corner (2006) concluded that: ‘‘the risk of aerosol transmission to cattle will be more pervasive than the risk of infection from ingestion’’. Although the effect of exposure was not constant over time, for herds in groups 2 and 3, after 5 years at risk, the
original hazard ratios increased to 0.72 and 0.51, respectively. It was estimated that it would take 8 and 13 years for herds in groups 2 and 3, respectively to equal the hazard of herds in group 0. In conclusion our study shows that targeted badger removal had a beneficial impact on the survival time to future BTB episodes in herds in and around areas where badgers were removed in county Laois. Acknowledgment We would like to thank William Sears, from the Department of Population Medicine, OVC, University of Guelph, for his contribution with data management. References Allison, P.D., 1995. Survival Analysis Using the SAS System: A Practical Guide. SAS Institute Inc., Cary, NC, p. 292. Corner, L.A., 2006. The role of wild animal populations in the epidemiology of tuberculosis in domestic animals: How to assess the risk. Veterinary Microbiology 112, 303–312. Corner, L.A., O’Meara, D., Costello, E., Gormley, E., 2005. Tuberculosis in Badgers: True Prevalence, Diagnostic Methods and Epidemiology (poster). Society for Veterinary Epidemiology and Preventive Medicine, Nairn, Scotland. Dohoo, I., Martin, S.W., Stryhn, H., 2003. Veterinary Epidemiology Research. Atlantic Veterinary College Inc., Charlottetown, PEI, Canada, 706 pp. Dolan, L.A., 1993. Badgers and bovine tuberculosis in Ireland: a review. In: Hayden, T.H. (Ed.), The Badger. Royal Irish Academy, Dublin, pp. 108– 116. Donnelly, C.A., Woodroffe, R., Cox, D.R., Bourne, J., Gettinby, G., Le Fevre, A.M., Mclnerney, J.P., Morrison, W.I., 2003. Impact of localized badger culling on tuberculosis incidence in British cattle. Nature 426, 834– 837. Eves, J.A., 1999. Impact of badger removal on bovine tuberculosis in east county Offaly. Irish Veterinary Journal 52, 199–203. Grambsch, P.M., Therneau, T.M., 1994. Proportional hazards test and diagnostics based on weighted residuals. Biometrics 46, 93–102. Griffin, J.M., Williams, D.H., Kelly, G.E., Clegg, T.A., O’Boyle, I., Collins, J.D., More, S.J., 2005. The impact of badger removal on the control of tuberculosis in cattle herds in Ireland. Preventive Veterinary Medicine 67, 237–266. Martin, S.W., Eves, J.A., Dolan, L.A., Hammond, R.F., Griffin, J.M., Collins, J.D., Shoukri, M.M., 1997. The association between the bovine tuberculosis status of herds in the east Offaly project area, and the distance to badgers setts, 1988–1993. Preventive Veterinary Medicine 31, 113–125. Monaghan, M.L., Doherty, M.L., Collins, J.D., Kazda, J.F., Quinn, P.J., 1994. The tuberculin test. Veterinary Microbiology 40, 111–124. More, S., 2007. Personal Communications. O’Connor, R., O’Malley, E., 1989. Badgers and Bovine Tuberculosis in Ireland. Economic and Research Institute, Burlington Road, Dublin. O’Keeffe, J.J., 2007. Personal Communications. O’Mairtin, D., Williams, D.H., Griffin, J.M., Dolan, L.A., Eves, J.A., 1998. The effect of a badger removal programme on the incidence of tuberculosis in a Irish cattle population. Preventive Veterinary Medicine 34, 47–56. Olea-Popelka, F.J., White, P.W., Collins, J.D., O’Keeffe, J.O., Kelton, D.F., Martin, S.W., 2004. Breakdown severity during a bovine tuberculosis episode as a predictor of future herd breakdowns in Ireland. Preventive Veterinary Medicine 63, 163–172.
Contents lists available at ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Targeted badger removal and the subsequent risk of bovine tuberculosis in cattle herds in county Laois, Ireland F.J. Olea-Popelka a,e,*, P. Fitzgerald b, P. White c, G. McGrath c, J.D. Collins c, J. O’Keeffe c,d, D.F. Kelton a, O. Berke a, S. More c, S.W. Martin a a
Department of Population Medicine, Ontario Veterinary College, University of Guelph, Clinical Research Building, Guelph, Ont., Canada N1G 2W1 District Veterinary Office County Laois, Ireland Centre of Veterinary Epidemiology and Risk Analysis, University College Dublin, Belfield, Ireland d Department of Agriculture, Fisheries and Food, Agriculture House, Kildare Street, Dublin 2, Ireland e Animal Population Health Institute, Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins 80523, USA b c
A R T I C L E I N F O
A B S T R A C T
Article history: Received 24 July 2007 Received in revised form 3 September 2008 Accepted 8 September 2008
We investigated the impact of targeted removal of badgers on the subsequent bovine tuberculosis (BTB) risk in cattle herds in county Laois, Ireland. The study period was 1989– 2005. For each of 122 targeted badger-removal licenses (permit to remove badgers in the proximity of cattle herds undergoing a serious BTB episode), the herd number (index herd) for which the license was given was obtained. The herds in the proximity of the index herd were identified from another database. The main ‘‘exposure’’ in our study was the geographical location of herds relative to the area in which targeted badger removal was conducted. We categorized herds into five different exposure groups: herds were classified as non-exposed and denoted as group 0 (reference group) if they were located 500 m or more from the edge of any parcel of land of the index herd; group 1, was the index herds, group 2 the immediate (contiguous) neighbors of the index herd, group 3 herds were not immediate neighbors but within 150 m and group 4 herds were between 150 m and 500 m distance from the edge of any parcel of land of the index herd, respectively. We conducted a survival analysis (allowing multiple failures per herd) to compare the hazard of having a BTB episode in any of the four groups of exposed herds vs. the hazard in herds in the reference group. We controlled for other known risk factors as well taking into account a temporal component. Our analysis showed that the hazard ratio for the index herds (group 1) were non-significantly increased, indicating that there was no difference in the hazard of failing a BTB test (after the targeted badger removal was conducted) between index herds and reference herds. For the rest of the herds farther away from badger removal activities the hazards were lower than herds in areas not under badger removal. The hazard in the reference group decreased over the study period. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Bovine tuberculosis Badger removal Ireland
1. Introduction
* Corresponding author at: Animal Population Health Institute, Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins 80523, USA. Tel.: +1 970 2975064; fax: +1 970 4912250. E-mail address: [email protected] (F.J. Olea-Popelka). 0167-5877/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.prevetmed.2008.09.008
One of the current constraints to the eradication of bovine tuberculosis (BTB) in Ireland is the existence of an important wildlife reservoir for Mycobacterium bovis (M. bovis); namely the badger (Meles meles). In Ireland, several studies have provided evidence that tuberculous badgers
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
are associated with M. bovis infection in cattle (O’Connor and O’Malley, 1989; Dolan, 1993; Martin et al., 1997). In addition, two field trials known as the ‘‘East Offaly’’ and ‘‘Four Area’’ projects (O’Mairtin et al., 1998; Eves, 1999; Griffin et al., 2005) have shown that intensive badger removal lead to a decrease in the subsequent levels of tuberculosis in cattle herds in the areas in which the badgers were removed compared to areas with restricted targeted, or no, removal of badgers. Since the early 1990s, targeted badger removal (also known as ‘‘reactive’’ badger culling) has been conducted through Ireland. This practice is initiated when a veterinary inspector, in the course of an epidemiological investigation of a BTB breakdown strongly suspects badgers as the source of infection. After obtaining a license to capture badgers, from the National Parks and Wildlife Service of the Department of the Environment, Heritage and Local Government, an area of 1–2 km radius around the affected farm is surveyed and badger removal takes place until few badgers remain and/or the BTB problem in the area has been cleared. Despite the potential large size of the removal area, usually only those setts on or close to the index herd have badgers removed. This activity differs from the two badger-removal projects mentioned above in the sense that badgers are removed only around herds with a serious BTB problem (i.e. two or more cattle deemed as ‘‘standard reactors’’ and one animal with a BTB lesion). Reactive culling is a response to the problem in the area and is an attempt to prevent the spread of BTB from badgers to other herds in the vicinity of the affected (index) herd. The objective of our study was to assess the impact of ‘‘targeted’’ badger removal on the subsequent BTB risk of herds in one county, Laois, in Ireland. For this, we conducted a survival analysis in which the outcome was ‘‘time to a new breakdown of BTB’’. Herds were classified as having different levels of exposure to badgers in the removal area based on their geographical location relative to the index herd. We tested the hypothesis that the hazard of BTB for herds in areas in which targeted badger removal was conducted, and in surrounding areas, does not differ from the hazard of BTB for herds in areas in which targeted badger removal was not conducted. 2. Material and methods The study was conducted in county Laois, Ireland. County Laois is a centrally located inland county, 90 km southwest of the capital city Dublin. All badger removal licenses files were collected for the period 1990–2005 from the Laois District Veterinary Office (DVO). Data on the herd (index herd) for which the license was issued, the date at which the license was opened, the date of the last badger removal (end of the license), the number of badgers removed, and the number of tuberculous badgers removed per license was entered into a Excel spread sheet (Microsoft1 Office, 2003). If a license was reopened we included this as a new license. The herd BTB history data were obtained from the Centre for Veterinary Epidemiology and Risk Assessment at the University College Dublin, Ireland. This centre
179
maintains computerized data on all herd-test results for BTB since January 1989. In Ireland, all herds are tested at least once annually; herds suspected of having BTB, or herds with BTB, are tested more frequently. Data were obtained for all herds in county Laois between the first test record (on or after 1 January 1989) until the last test record during the year 2005. 2.1. Exclusion criteria All herds containing fewer than seven animals were excluded from our study. Also, herds that ceased their activities before 2003 were not included in the study. This exclusion was necessary because the computerized ‘‘neighboring’’ file (see below) which was essential for our study contained data only on herds that were active during 2003. 2.2. Exposure categories Herd exposure was based on their location relative to the index herd (and hence relative to the activity of targeted badger removal). Using data from the ‘‘neighboring’’ file created with a Geographical Information System (GIS), herds that were located in areas 500 m or more from the edge of any parcel of land of the index herd at the time of the BTB breakdown were classified as non-exposed and denoted as group 0 (reference group). Herds that originated a license were identified as the index-herds (group = 1). All the immediate neighbors (<25 m distance from the edge of any parcels of land of the index herd) were identified and classified as group 2 herds. Herds at distances of 25 < 150 m and 150 < 500 m distances of the edge of any parcel of land of the index-herd at the time of the BTB breakdown were identified as exposure groups 3 and 4, respectively. Considering that typical Irish farms are composed of several parcel of land, and each parcel of land varies in their dimensions, herds recorded at more than 500 m of any of the edge of the index farm could be located at distances as large or larger than 3 km (O’Keeffe, 2007 personal communication). Thus, the distance measures used to create the exposure categories are ordered but the cutpoints do not reflect the actual distance from the badger removal areas. Herd classification was allowed to change over time as explained below. 2.3. Exposure over time The starting point to accumulate time at risk was the first recorded herd test in 1989, or the first herd-test of new herds after that date. If the herd was positive to BTB at its first test, we used the date at which the herd was declared ‘‘clear’’ (i.e. after the herd had passed three consecutive negative herd-tests, usually a period of approximately 10 months) to begin accumulating time-at-risk. Assuming the herd was not exposed to badger removal at its first test, herds accumulated time at risk in exposure group 0 until one of the following scenarios occurred: the herd experienced a change in exposure status (to one of the groups 1–4) (Fig. 1), or
180
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
Fig. 1. Exposure over time: change in exposure categories for herds in county Laois, Ireland after the targeted badger removal was ended, 1989–2005.
Fig. 2. (a) Exposure over time: example of a herd tested for the last time (censored time) without any events during the study period in county Laois, Ireland 1989–2005. (b) Exposure over time: example of the progress herds could have from what we considered as minimally exposed (group 4) to higher levels of exposure (groups 3, 2, or 1) for the duration of the study 1989–2005 in county Laois, Ireland.
the herd experienced a BTB episode (failure time) (Figs. 1 and 3), or the herd was tested for the last time (censored time) (Fig. 2a). When a herd became an index herd (group 1), it’s exposure classification remained as a group 1 for the entire duration of our study. The classification of immediate neighbors (group 2) did not change unless they became an index herd. Herds in groups 3 or 4 only changed their status if they subsequently were located closer to another index herd (e.g. from being >150–500 m. distant from an index herd (group 4), to being an immediate neighbor for another index herd (group 2). Herds could only progress from what we considered as minimally exposed (group 4) to higher levels of exposure (groups 3, 2, or 1) for the duration of the study (Fig. 2b). With the exception of the reference group, exposure to the badger removal activities (group 1 through 4) was deemed to begin when badger removal activities ended. This rule was invoked because badger licenses often were kept open until new tuberculosis problems in the groups 1 and 2 herds had been overcome.
2.4. Outcome The outcome of experiencing a herd breakdown with tuberculosis denoted a herd having at least one or more standard reactors1 at the Single Intradermal Comparative Tuberculin Test (SICTT). If a herd developed a BTB episode, it was removed from the at-risk group in its existing exposure category until it had passed three negative herd tests. Thereafter, it was deemed to be at risk of another BTB breakdown; thus a herd could have multiple failures (BTB episodes) during the study period (Fig. 3 and Table 2). The cumulative exposure time in each exposure category was calculated until the herd broke down with BTB (Fig. 3). The censoring time was the date of the last recorded herd test before 1 January 2006.
1 In Ireland, a standard reactor is recorded when the bovine reaction is >4 mm larger than the avian reaction, or if local clinical signs (such as edema, exudation, necrosis or pain) are present at the site of injection of bovine tuberculin (Monaghan et al., 1994).
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
181
Fig. 3. Outcome over time: example of a herd having multiple failures (BTB episodes) during the study period 1989–2005 in county Laois, Ireland.
2.5. Statistical analysis SAS V8.2 (Statistical Analytical System Institute Inc., Cary, NC, USA), was used for data manipulation, and descriptive analysis. STATA (Stata Statistical Software: Release 9 College Station, TX, USA) was used for further data manipulation and survival analysis. Univariable and multivariable Cox’s proportionalhazard models were used to contrast the hazards for a BTB breakdown in each of the four exposure groups relative to the hazard in the non-exposed group (reference group) using STCOX in STATA. We included the herd size, herd type, the length of the badger removal period, the number of badgers and tuberculous badgers removed and the time (year) in which the removal activities took place as covariates. We accounted for clustering (repeated measurements within a herd) by using the ROBUST standard error option. We checked the proportional-hazard assumption by assessing, visually, if the log( log) survival lines were parallel for the different exposure groups (Allison, 1995). More formally, we assessed if the hazard-ratio varied over time by analyzing the x2 Shoenfeld residuals (Grambsch and Therneau, 1994), and if significant (p < 0.05), we included the exposure as a time varying covariate (Dohoo et al., 2003). Exposure time and time-to-failure were measured in months. Interaction terms between the potential confounders and the exposure groups were evaluated. 3. Results 3.1. Descriptive Over the 16-year study period, 135 badger removal licenses were granted; complete data were available for 122 of these. The data for nine of the omitted licenses were excluded because of the absence of the index herd in the 2003 contiguous file, making it impossible for us to locate their neighbors; four others for which the end date of the licence was not recorded were also omitted. In total, 1257 badgers were removed and 15% of these (n = 182) were deemed to have BTB lesions at the gross post-mortem examination. Table 1 shows the distribution of badgers and tuberculous badgers removed at the completion of each license by year. There was no statistically significant
difference in the annual risk of badgers with a gross BTB lesion (x2 = 13.0, d.f. = 12, p = 0.36). Neither was there a significant linear trend on the levels of badger BTB over time using the Cochran-Armitage test for trend (x2 = 1.531, d.f. = 1, p-value = 0.216). On average 3645 herds were tested each year. We included only data on 2904 herds after implementing our exclusion criteria. Among these herds 55% did not have a BTB breakdown over the study period. The distribution of number of BTB breakdowns per herd is shown in Table 2. Among herds that had BTB, 36% had repeated episodes; the distribution of occurrences fitted a negative binomial but not a Poisson distribution. The 1st quartile, median and 3rd quartile for the duration of the BTB episode were 11, 12 and 15 months, respectively. Table 1 Number of badgers and tuberculous badgers removed by year in county Laois, Ireland 1991–2004. Year
No. of badgers
No. of tuberculous badgers
% Positive
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
136 62 18 48 59 202 114 159 34 69 114 161 59 22
21 10 1 6 6 34 24 25 3 6 18 24 4 0
15.4 16.1 5.6 12.5 10.2 16.8 21.1 15.7 8.8 8.7 15.8 14.9 6.8 0.0
Table 2 Number of BTB episodes per herd in county Laois, Ireland 1989–2005. Number of BTB breakdowns
Number of herds
0 1 2 3 4 5
1590 846 325 111 28 4
Total
2904
Percentage 54.8 29.1 11.2 3.8 1.0 0.1 100
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
182
Table 3 Distribution of time at risk (in months) by exposure category for herds in county Laois, Ireland 1989–2005. Percentiles
Exposure categorya 0
1
2
3
4
Smallest 1% 25% 50% 75% 99%
0.2 3.8 35.4 83.4 156 202.8
0.2 0.3 24.8 53 94.1 172
0.03 1.3 37.2 66.7 108.6 154.8
0.1 0.9 36.7 61.2 104.2 176.7
0.03 0.8 34.3 60.8 103.8 174.8
Largest
208.1
178
188.2
190.5
181.3
a
See Section (2.2) for description of each exposure category.
hazard for the index herds did not differ from the referent. The hazard ratio was not constant for herds in groups 2 and 3 (p 0.001 and p = 0.043, respectively); the hazard ratio for these groups was initially low, but increased slightly over the years while remaining low relative to the hazard for herds in group 0. Exposure group 4 herds had a lower risk of a BTB breakdown than the referent group. As the number of animals tested in a herd increased by 10, so did the hazard of failing a BTB herd test (HR = 1.05, p < 0.001). The highest hazard for year in 1990 (HR = 2.2, p < 0.001) significantly decreased over time until the end of 1995. The hazards between 1996 and 2003 did not differ from that in 1989 (data not shown). 4. Discussion
Historical records for county Laois showed that the herd incidence (% of new herds restrictions) over the 16-year study period decreased beginning in 1991 and again in 1996 but retained a constant level from 1999 onwards. The distribution of time at risk (in months) for each exposure category is shown in Table 3; not surprisingly, exposure category ‘‘0’’ had the longest median (or mean) values due to the time sequence of initiating badger removal licenses during the study period. 3.2. Survival analysis All the exposure categories and confounders had a significant association with the time to fail a BTB test in a simple univariate survival model (Table 4). The results for the main risk factor, exposure category, indicated that group 1 herds had a higher hazard than group 0 herds. Conversely, exposure groups 2, 3 and 4 had lower hazard ratios indicating longer survival. As the number of badgers and tuberculous badgers removed increased, the hazard of a herd failing a BTB test decreased. As the number of animals tested in a herd increased, so did the hazard of failing a herd test. The hazard for beef and suckler2 herds was lower than dairy herds. The hazard of failing a herdtest decreased over the study period. After controlling for the other variables in the model, herd type, the effect of number of badgers removed and tuberculous badgers removed, was no longer significant. The log( log of the survival probability) graph for a BTB breakdown, by exposure category, revealed that the survival lines were not parallel, visually, indicating that the hazard ratio for some exposure categories varied significantly over time. The Shoenfeld residuals test also indicated that our model violated the proportionality assumption (x2 = 212.8, d.f. = 22, p < 0.001). Thus, we controlled and adjusted for this by including the exposure categories as a time varying covariate (Dohoo et al., 2003). The final multivariable survival model is shown in Table 5. We controlled for year of study and herd size. The
2 Breeding cows from the beef type breeds (Charolais, Limousine, Hereford, Aberdeen Angus, etc.) crossed with a beef breed type bull to produce animals suitable for production of high-quality beef. Calves when born run with their dam for 4–8 months. A suckler herd normally does not produce a finished beef animal, but rather a yearling animal for further feeding.
Since 2003, a considerable amount of controversy has surrounded the issue of badger removal as a measure for controlling BTB in cattle. The East Offaly and Four Area projects (Eves, 1999; Griffin et al., 2005) compared areas in which extensive proactive badger removal was conducted to eliminate badgers as a potential source of M. bovis infection for cattle, and this resulted in a decrease of the BTB levels in cattle relative to those where minimal (restricted targeted), or no, badger removal took place (Eves, 1999; Griffin et al., 2005). In contrast, in the randomized badger culling trial (RBCT) in southwest England, areas with ‘‘reactive’’ badger culling had an increased incidence of BTB in cattle (Donnelly et al., 2003) and herds in areas around the extensive badger removal areas also had an increased incidence of BTB. Badger social disruption was suggested as the main reason for the observed increased incidence of BTB in cattle-herds in these areas in England. In our study, the protective hazard ratios obtained for groups 2–4 (0.42, 0.33 and 0.48, respectively) indicate that herds in areas around the targeted removal activities had a longer survival time to the next BTB episode compared to herds outside of these areas (reference group). One could argue that, the lower hazard for the exposed groups (2–4) could be an ‘‘artifact’’ due to an increase of BTB in the nonexposed group more than a decrease of BTB in the exposed groups. We investigated this aspect, but found no evidence of an increase over time of the BTB levels in herds in areas in which badgers were not removed. More specifically, we observed a statistically significant decrease of the hazard among herds in the reference group over the first 6 years of the study, after which it remained constant (data not shown). Thus, the lower hazard ratios for herds in groups 2–4 suggest a true decrease of the risk of BTB among these herds. With regards to potential misclassification bias in our outcome variable (BTB event), we recognize that the SICTT is not a perfect test. As a reasonable guide, the SICTT as administered in Ireland has a specificity (proportion of non-infected cattle that test negatively) of 99.8–99.9% and a sensitivity (proportion of infected cattle that test positively) of between 68% and 95% (Monaghan et al., 1994). Thus, it is likely that at the herd test, not all animals infected with M. bovis are detected. Nonetheless, herd sensitivity is higher than the animal-level sensitivity and
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
183
Table 4 Univariate hazard ratios, of future bovine tuberculosis herd breakdown after targeted badger removal took place in county Laois, Ireland 1989–2005. Variables
Hazard ratio
Robust S.E.
P-value
95% HR CL
Reference 1.62 0.71 0.48 0.57
– 0.18 0.05 0.08 0.05
– <0.0001 <0.0001 <0.0001 <0.0001
– 1.29, 0.63, 0.36, 0.48,
a
1. Badger removal exposure category 0 1 2 3 4
2.04 0.82 0.63 0.69
2. Number of badgers removed (increase in 1 badger)
0.97
0.005
<0.0001
0.97, 0.98
3. Number of tuberculous badgers removed (increase in 1 positive badger)
0.90
0.02
<0.0001
0.85, 0.94
4. Herd size (Increase in 10 cattle)
1.05
0.008
<0.0001
1.04, 1.07
5. Length of removal (months)
0.99
0.0001
0.018
0.99, 0.99
6. Year of study period (increase in 1 year)
0.97
0.005
<0.0001
0.96, 0.99
7. Herd type Beef Suckler
0.66 0.72
0.046 0.043
<0.0001 <0.0001
0.57, 0.75 0.64, 0.81
a
See Section (2.2) for description of each exposure category.
Table 5 Multivariable hazard ratios by exposure group, of future bovine tuberculosis herd breakdown after targeted badger removal took place in county Laois, Ireland 1989–2005. Variables
Hazard ratio
Robust S.E.
P-value
95% HR CL
Badger removal exposure category 0 1 2 3 4
Reference 1.15 0.42 0.33 0.48
– 0.25 0.06 0.08 0.08
– 0.51 <0.0001 <0.0001 <0.0001
– 0.74, 0.33, 0.21, 0.35,
Covariates Herd size (inc. in 10 animals)
1.05
0.007
<0.0001
1.04, 1.07
Time varying covariates (by month) 1 2 3 4
1.004 1.009 1.007 1.004
0.0037 0.0017 0.0035 0.0022
0.32 <0.0001 0.043 0.107
0.99, 1.01, 1.00, 0.99,
a
a
1.78 0.55 0.54 0.65
1.01 1.01 1.01 1.01
See Section (2.2) for description of each exposure category.
should be non-differential with respect to major exposure factors. If a herd had a BTB event during the study period, we considered that herd as ‘‘clear’’ (hence, cumulating ‘‘time at risk’’ again) only if the herd tested negatively on at least three consecutive tests (2 reactor re-test plus one ‘‘6month’’ check test). This was a ‘‘conservative’’ approach and should minimize any bias. In this study, we focused on the herd-risk of BTB in areas with ‘‘targeted’’ badger removal, analogous to those areas described as ‘‘reactive’’ culling in other studies, in comparison to the herd-risk in areas that were relatively distant from badger removal activities. We selected county Laois for our study because of the availability of welldocumented data on targeted badger removal over a considerable period of time (since 1989 onwards). Historical data show that county Laois over the last 10 years has experienced a decrease in BTB levels. In our study, the percentage of tuberculous badgers was obtained only by post-mortem BTB gross lesion findings. The percentage of tuberculous badgers removed did not differ over the study period. We acknowledge that the
post-mortem diagnosis it is not a sensitive method for detecting tuberculosis. Corner et al., 2005 have described the true level of BTB infection among badger population with a more sensitive methodology for identifying infection. The results of such studies have shown that approximately 44% of badgers are infected with M. bovis. The lower sensitivity in our study should be nondifferential by exposure category, but could reduce the significance of this specific variable. Indeed, the number of tuberculous badgers removed was not a predictor of future hazard of BTB in cattle herds. The assigning of a badger-removal exposure category to a herd was done by considering the end date for each badger removal license as a starting point of the time at risk for each exposure category. We did this, keeping in mind that badgers tended to be removed around groups 1 and 2 herds until the BTB problem in cattle herds had been cleared. During this time, these herds often were tested more frequently than herds in other areas, and hence would have had an apparent (if not real) increased risk of BTB.
184
F.J. Olea-Popelka et al. / Preventive Veterinary Medicine 88 (2009) 178–184
In our final model, the non-significant hazard ratio obtained for the index herds (group 1), indicated that there was no difference in the hazard of failing a BTB test (after the targeted badger removal was conducted) between index herds and herds in the reference group (at considerable distances from badger removal activities). Previous research indicated that the group of index herds would be classified as a future ‘‘high’’ risk group of herds, considering they recently had a serious BTB episode (Olea-Popelka et al., 2004). Data on badgers, and badger removal, were not available for that study. In the current study, herds that had a severe BTB breakdown did not have a higher risk of failing a future BTB test after the badger removal was completed in the area suggesting a beneficial impact of targeted removal on the survival time in these herds. The index herd and its immediate neighbors undergo a more frequent testing regime than other herds. As conducted in Ireland, this strategy increases the probability of detecting infected animals (therefore herds) and together with the removal of badgers, decreases the reservoirs of infection for other cattle (herds) in the area. For this reason, we decided to accumulate time at risk for each exposure category only after the badger removal activities ceased. In addition, we assumed only a monotonic inverse association of exposure category with BTB risk and did not allow for other possibilities. We consider it important to understand the impact of badger removal activities on the dynamics of badger populations, and to distinguish between social disruption of badger groups and the effects of removal on levels of BTB in cattle herds. Alternative analyses of the data to evaluate these assumptions are planned in the future. Any causal link between reactive badger removal and bovine tuberculosis in cattle herds is certain to require a time lag sufficient to enable the hypothesized cascade of events to be completed including perturbation in disturbed badger populations, increased movement and contact between badgers within and without the same social groups, increased contact between badgers and cattle, infection of cattle, the development of responsiveness to tuberculin following establishment of infection in cattle under natural conditions, the scheduling of the (annual) herd test, the inspection of reactor animals at slaughter and laboratory confirmation of disease (More, 2007 personal communication). The risks posed by tuberculous badger to cattle have being recently described and analyzed by Corner, 2006. After considering different factors, Corner (2006) concluded that: ‘‘the risk of aerosol transmission to cattle will be more pervasive than the risk of infection from ingestion’’. Although the effect of exposure was not constant over time, for herds in groups 2 and 3, after 5 years at risk, the
original hazard ratios increased to 0.72 and 0.51, respectively. It was estimated that it would take 8 and 13 years for herds in groups 2 and 3, respectively to equal the hazard of herds in group 0. In conclusion our study shows that targeted badger removal had a beneficial impact on the survival time to future BTB episodes in herds in and around areas where badgers were removed in county Laois. Acknowledgment We would like to thank William Sears, from the Department of Population Medicine, OVC, University of Guelph, for his contribution with data management. References Allison, P.D., 1995. Survival Analysis Using the SAS System: A Practical Guide. SAS Institute Inc., Cary, NC, p. 292. Corner, L.A., 2006. The role of wild animal populations in the epidemiology of tuberculosis in domestic animals: How to assess the risk. Veterinary Microbiology 112, 303–312. Corner, L.A., O’Meara, D., Costello, E., Gormley, E., 2005. Tuberculosis in Badgers: True Prevalence, Diagnostic Methods and Epidemiology (poster). Society for Veterinary Epidemiology and Preventive Medicine, Nairn, Scotland. Dohoo, I., Martin, S.W., Stryhn, H., 2003. Veterinary Epidemiology Research. Atlantic Veterinary College Inc., Charlottetown, PEI, Canada, 706 pp. Dolan, L.A., 1993. Badgers and bovine tuberculosis in Ireland: a review. In: Hayden, T.H. (Ed.), The Badger. Royal Irish Academy, Dublin, pp. 108– 116. Donnelly, C.A., Woodroffe, R., Cox, D.R., Bourne, J., Gettinby, G., Le Fevre, A.M., Mclnerney, J.P., Morrison, W.I., 2003. Impact of localized badger culling on tuberculosis incidence in British cattle. Nature 426, 834– 837. Eves, J.A., 1999. Impact of badger removal on bovine tuberculosis in east county Offaly. Irish Veterinary Journal 52, 199–203. Grambsch, P.M., Therneau, T.M., 1994. Proportional hazards test and diagnostics based on weighted residuals. Biometrics 46, 93–102. Griffin, J.M., Williams, D.H., Kelly, G.E., Clegg, T.A., O’Boyle, I., Collins, J.D., More, S.J., 2005. The impact of badger removal on the control of tuberculosis in cattle herds in Ireland. Preventive Veterinary Medicine 67, 237–266. Martin, S.W., Eves, J.A., Dolan, L.A., Hammond, R.F., Griffin, J.M., Collins, J.D., Shoukri, M.M., 1997. The association between the bovine tuberculosis status of herds in the east Offaly project area, and the distance to badgers setts, 1988–1993. Preventive Veterinary Medicine 31, 113–125. Monaghan, M.L., Doherty, M.L., Collins, J.D., Kazda, J.F., Quinn, P.J., 1994. The tuberculin test. Veterinary Microbiology 40, 111–124. More, S., 2007. Personal Communications. O’Connor, R., O’Malley, E., 1989. Badgers and Bovine Tuberculosis in Ireland. Economic and Research Institute, Burlington Road, Dublin. O’Keeffe, J.J., 2007. Personal Communications. O’Mairtin, D., Williams, D.H., Griffin, J.M., Dolan, L.A., Eves, J.A., 1998. The effect of a badger removal programme on the incidence of tuberculosis in a Irish cattle population. Preventive Veterinary Medicine 34, 47–56. Olea-Popelka, F.J., White, P.W., Collins, J.D., O’Keeffe, J.O., Kelton, D.F., Martin, S.W., 2004. Breakdown severity during a bovine tuberculosis episode as a predictor of future herd breakdowns in Ireland. Preventive Veterinary Medicine 63, 163–172.